Optimization with a twist...

blackmore 2011.10.16 09:34 #1

What theory?

mbirrell 2011.10.16 12:00 #2

That some combination of time frames will work better than others. Example: if I was to use a M5 time frame and and M1 time frame with Stochastic it may not have as high profit ratio as say M30 and M5.

I know there is a difference in profit ratio because I have tried it. I have tested 6 different time frame combinations and some do very well - some don't.

I am going to test the top 3 patterns out - it will take me quite a while to do this as I have to back test each pattern before I can put all the numbers into my EA. I hope to cut the testing time down as the results have shown around 40 patterns have a profit ratio above 2.4 - and that's with unoptimized variables.

How to combine result Looking for patterns Please help me solve

William Roeder 2011.10.16 12:21 #3

mbirrell:
Has anyone tried using optimization to figure out which combination of indicators and time frames work best?

There are thousands of indicators. A thousand factorial is far larger than the number of atoms in the universe. You're probably going to need this.

blackmore 2011.10.16 12:39 #4

mbirrell:

That some combination of time frames will work better than others. Example: if I was to use a M5 time frame and and M1 time frame with Stochastic it may not have as high profit ratio as say M30 and M5.

I know there is a difference in profit ratio because I have tried it. I have tested 6 different time frame combinations and some do very well - some don't.

I am going to test the top 3 patterns out - it will take me quite a while to do this as I have to back test each pattern before I can put all the numbers into my EA. I hope to cut the testing time down as the results have shown around 40 patterns have a profit ratio above 2.4 - and that's with unoptimized variables.

I see.

You might like to check the ATC news articles --> https://championship.mql5.com/2011/en/news/ such as this one. https://championship.mql5.com/2011/en/news/86

Pretty interesting discoveries about TFs. As well as others, related.

I don't use oscillators--- but I heard before of their logical interpretations behind to work better for H1 & D1, exception is H4 and lower TF. anyway, I am definitely the wrong candidate for this.

I rather work on Price Action, Price-Based and Levels theories rather than this because that is easier for me.

Please cast your vote Neural Networks Predicting the next candle

mbirrell 2011.10.16 20:50 #5

WHRoeder:
There are thousands of indicators. A thousand factorial is far larger than the number of atoms in the universe. You're probably going to need this.

yes that is true - if you single out just a few that you understand well like I have you will get interesting results as I have.

mbirrell 2011.10.18 11:53 #6

one of thousands of indicators not to mention combinations achieved this.... Graph is over last 1.5 months. Might make money or might lose money - I am about to find out.